Rowe Jessica, Nogovitsyn Nikita, Mazurka Raegan, Squires Scott D, Hassel Stefanie, Poppenk Jordan, Dunlop Katharine, Zamyadi Mojdeh, Milev Roumen V, Foster Jane A, Arnott Stephen R, Lam Raymond W, Uher Rudolf, Rotzinger Susan, Kennedy Sidney H, Frey Benicio N, Harkness Kate L
Department of Psychology, Queen's University, Kingston, Ontario, Canada.
Mood Disorders Treatment and Research Centre, St Joseph's Healthcare, Hamilton, Ontario, Canada.
JAMA Netw Open. 2025 Aug 1;8(8):e2525147. doi: 10.1001/jamanetworkopen.2025.25147.
The limited success of major depressive disorder (MDD) treatments is largely due to the disorder's etiological and pathophysiological heterogeneity. Addressing this heterogeneity is essential for developing accurate prognostic models and personalized treatment strategies.
To characterize MDD heterogeneity using a mechanism-first latent profile analysis based on environmental, neurostructural, and neurofunctional indicators, and to validate profiles via associations with MDD course, severity, and antidepressant treatment remission.
DESIGN, SETTING, AND PARTICIPANTS: This cross-sectional study used data from 2 Canadian Biomarker Integration Network in Depression (CAN-BIND) studies: CAN-BIND-1 (2014-2017), a multicenter outpatient antidepressant trial, and CAN-BIND-4 (2015-2018), a single-site study. Data analyses were completed from February to September 2024. Participants meeting Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition) diagnostic criteria for unipolar depression were included. Individuals with lifetime bipolar, psychotic, substance use disorder, acute suicidality, and neurological disorders were excluded.
In CAN-BIND-1, patients received 10 to 20 mg of escitalopram daily; nonresponders at 8 weeks received aripiprazole augmentation for 8 additional weeks. CAN-BIND-4 was observational.
Primary outcomes were latent profiles derived from childhood maltreatment (CM; semistructured interview); hippocampal, amygdala, thalamus structural volume (SV); anterior cingulate thickness (image segmentation); and DMN functional connectivity (average time series of the blood oxygen level-dependent signal). Secondary outcomes included associations with MDD course, symptom severity (including anhedonia, measured using Montgomery-Åsberg Depression Rating Scale), and remission rates.
In a sample of 309 adults with clinical depression (mean [SD] age, 33.81 [13.17] years; 206 female [66.67%]), 4 profiles emerged: (1) low CM and high SV, (2) low CM and low SV, (3) high CM and high SV, and (4) high CM and low SV with default mode network hypoconnectivity. Profile 4 was associated with the worst course, with the highest morbidity (mean number of years of morbidity, 19.91 years; 95% CI, 12.45-20.69 years), anhedonia (mean, 10.72; 95% CI, 9.74-11.70), and lowest remission rate (mean, 21.5%; 95% CI, 17.6%-23.5%) at week 16. Profile 3 had the highest remission rates (mean, 90.9%; 95% CI, 63.4%-118.0%).
In this cross-sectional study of 309 adults with depression, 4 latent profiles were identified. Default mode network hypoconnectivity defined profile 4, supporting its role as a key neural indicator of antidepressant response. CM was associated with both the highest and lowest remission rates, indicating it does not uniformly project negative outcomes and suggesting that neurobiological resilience in the context of childhood trauma may have contributed to more favorable clinical outcomes; further research is needed to refine clinical applications.
重度抑郁症（MDD）治疗的成效有限，很大程度上是由于该疾病在病因学和病理生理学上的异质性。解决这种异质性对于开发准确的预后模型和个性化治疗策略至关重要。
使用基于环境、神经结构和神经功能指标的机制优先潜在剖面分析来表征MDD的异质性，并通过与MDD病程、严重程度和抗抑郁治疗缓解情况的关联来验证剖面。
设计、设置和参与者：这项横断面研究使用了来自2项加拿大抑郁症生物标志物整合网络（CAN-BIND）研究的数据：CAN-BIND-1（2014 - 2017年），一项多中心门诊抗抑郁试验；以及CAN-BIND-4（2015 - 2018年），一项单中心研究。数据分析于2024年2月至9月完成。纳入符合《精神障碍诊断与统计手册》（第四版）单相抑郁症诊断标准的参与者。排除有双相情感障碍、精神病性障碍、物质使用障碍、急性自杀倾向和神经系统疾病终生史的个体。
在CAN-BIND-1中，患者每天接受10至20毫克艾司西酞普兰治疗；8周时无反应者额外接受8周阿立哌唑增效治疗。CAN-BIND-4为观察性研究。
主要结局是从童年虐待（CM；半结构化访谈）、海马体、杏仁核、丘脑结构体积（SV）、前扣带回厚度（图像分割）和默认模式网络功能连接性（血氧水平依赖信号的平均时间序列）得出的潜在剖面。次要结局包括与MDD病程、症状严重程度（包括使用蒙哥马利 - 阿斯伯格抑郁评定量表测量的快感缺失）和缓解率的关联。
在309名患有临床抑郁症的成年人样本中（平均[标准差]年龄，33.81[13.17]岁；206名女性[66.67%]），出现了4种剖面：（1）低CM和高SV，（2）低CM和低SV，（3）高CM和高SV，以及（4）高CM和低SV且默认模式网络连接性降低。剖面4与最差的病程相关，发病率最高（平均发病年数，19.91年；95%置信区间，12.45 - 20.69年），快感缺失最严重（平均，10.72；95%置信区间，9.74 - 11.70），且在第16周缓解率最低（平均，21.5%；95%置信区间，17.6% - 23.5%）。剖面3的缓解率最高（平均，90.9%；95%置信区间，63.4% - 118.0%）。
在这项对309名抑郁症成年人的横断面研究中，识别出了4种潜在剖面。默认模式网络连接性降低定义了剖面4，支持其作为抗抑郁反应关键神经指标的作用。CM与最高和最低缓解率均相关，表明它并非总是预示负面结果，提示童年创伤背景下的神经生物学复原力可能促成了更有利的临床结局；需要进一步研究以完善临床应用。
